At present, an Ethernet loop architecture is widely used in both metropolitan area network and Intranet to improve the reliability of the network. One loop corresponds physically to a loop-connected Ethernet topology, which is a group of Ethernet switchers connected with each other into a loop. The Ethernet loop protection protocol defines the role of each node in the loop respectively. A master node is a main decision-making and control node in the loop. There should be one and only one master node in each loop. All the other nodes except the master mode in the loop are transmission nodes. Each node uses a control Virtual Local Area Network (VLAN) to transmit protocol messages, and cooperates with each other to protect the loop, and uses a protection VLAN to transmit data messages such that the data messages and the control messages are separated from each other by different VLANs. When one link in the loop is disconnected, a loop protection mechanism is enabled to recover communication between each node in the loop network.
Each transmission node in the loop monitors its own port link state in real time. When a certain link in the loop fails, nodes at two ends of the link detect that the port is down, and a corresponding control protocol message should be sent immediately from another failure-free port to notify the master node of this change. After receiving this message, the master node invokes a forwarding data function of a slave port, and notifies each transmission node in the loop to refresh its own media access control (MAC) address list. When a node adjacent to a failed link detects link failure recovery, it will block immediately the data message forwarding function of the port just recovered and enter into a pre-forwarding state where the port can forward control protocol messages instead of protection VLAN data messages. At this time, a HELLO message which is sent by the master node periodically can be returned to the slave port of the master node through a control VLAN loop. Thus, the master node considers that the link is recovered, and re-blocks the protection VLAN data forwarding function of the slave port, and sends the corresponding control protocol message from a master port to notify the transmission nodes in the loop that the link is recovered and notify the transmission nodes to refresh their respective MAC address lists. A port which is in the pre-forwarding state receives this message and changes the pre-forwarding state to a forwarding state, while invoking the protection VLAN data forwarding function of the port.
In such a failure detection and recovery mechanism, the node adjacent to the failed link detects the recovery of the link, and immediately prevents the corresponding port from entering into the pre-forwarding state. After the master node perceives the recovery of the link, the control protocol message (which is used to notify the transmission nodes in the loop to refresh addresses and trigger state switching of the corresponding port) is sent by the master port. If this message is lost, the port adjacent to the recovered link will be always in the pre-forwarding state, namely, data messages can not be forwarded, and in the meanwhile the master node also blocks the data forwarding function of the slave port, thereby causing disconnection of the communication link and further resulting in the loss of the data messages.